Non-Normal Estimation of Multiple Spatial Patterns of Disease using Multivariate Skews Normal Process
Conference
64th ISI World Statistics Congress
Format: CPS Abstract
Session: CPS 10 - Disease and mortality modelling
Monday 17 July 8:30 a.m. - 9:40 a.m. (Canada/Eastern)
Abstract
Multivariate conditional autoregressive models based on the Gaussian are commonly applied in the analysis of multivariate spatial data. However, the modelled data could be highly tailed and skewed. We present, as an alternative, a multivariate skew-normal distribution in the analysis of multiple non-Gaussian spatial data. The estimation of the spatial patterns is fully Bayesian. Simulations and an application to estimate district HIV rates in South Africa are used for illustrating the capabilities of the proposed non-Gaussain approach to the analysis of multivariate skewed data.